Modeling and analysis of management for an agro-ecosystem using an agent-based model interface for the soil and water assessment tool (SWAT)

Poor water quality across the Mississippi River basin and its outlet, the Gulf of Mexico, is undermining the health of ecosystems, economies, and public health. Agricultural production in the watershed has been identified as the dominant factor contributing to poor water quality. Substantial investment by communities, governments, and research is dedicated to identifying appropriate agricultural management and practices to mitigate pollutants entering these waterbodies. Efforts must acknowledge diversity in agricultural production, stakeholders, environmental and societal factors to successfully address water quality issues. Consequently, it is important to develop comprehensive tools that can inform decision-makers with practical solutions with respect to environmental, economic, societal, and policy goals. In this study, a coupled human-natural systems model and software interface was developed to simulate feasible agricultural management and policy changes in an east-central Illinois watershed to identify strategies suitable for producers and policy-makers. The Soil and Water Assessment Tool (SWAT) was calibrated using publicly available sources and comparable previous studies for nutrient loads, water yield, tile-drained flow, and crop yields (natural-systems outcomes). SWAT modeling performance was satisfactory or better with respect to previous studies (annual PBIAS for nitrogen, phosphorous, water flow, and crop yields < 20%). An agent-based model was developed for community and farmer behavior to simulate hypothetical policy initiatives, economic returns, best management practice adoption (human-systems outcomes). The models were coupled to form a software interface, ITEEPGAM (the Integrated Tool for Environmental Economic and Policy Goals in Agricultural Management). ITEEPGAM was used to perform an analysis of watershed-specific BMPs (winter cover cropping, nutrient application timing, and drainage water management) along with fertilizer reductions and hypothetical tax, incentive, cost share policy initiatives. The development of ITEEPGAM and scenario analysis demonstrated that significant and complex natural systems and human systems phenomenon can be satisfactorily modeled and analyzed for potentially greater environmental and economic gains. The study showed a lower potential for environmental gains (8%-10% reductions in nitrogen and phosphorous) that other BMP studies in similar areas due to a smaller set of BMPs considered and an incorporation of an agent-based model to drive adoption behavior. Modeling results and agent behavior highlighted the importance of agent profiles, focusing input ranges and practical management choices to achieve useful conclusions. In this study, it was evident that enforcing fertilizer reductions beyond 15% were impractical for farmers. The scenario analysis highlighted effective policy instruments and potential redundancies. Incentives presented the most cost-effective return for designing community policy, but were not suitable to budgets beyond $1,000,000 as incentives served to supplement farmer returns without environmental benefit. Cost shares were effective at increasing adoption, but only up to a threshold of adopters. Small tax schemes could promote adoption and generate revenue for the community. Winter cover cropping coupled with small fertilizer reductions with the greatest potential for preserving economic performance and improving environmental gains while maintaining adoption rates. In the case of nutrient management paired with fertilization reductions, it could only offset very small fertilizer reductions and was therefore not economical